Traditionally, fusion plasma physics has applied a ‘divide-and-conquer’ approach to separately investigate various key processes like turbulent transport, large-scale instabilities, and energetic particle dynamics. In recent years, it has become increasingly clear, however, that these processes are all interlinked in a nonlinear way, pointing to the need to better understand their mutual interactions. In this context, first-principles based models are being developed and applied, but they tend to lead to enormously computationally expensive simulations. While the latter can provide valuable insights, their use for many important applications, from the detailed analysis of existing experimental data to the design optimization of future devices to real-time control of plasma experiments are clearly out of reach. Reduced models, on the other hand, are much faster and have been the mainstay of whole device modeling for decades. However, these simpler models are often based on crude assumptions and sometimes are even inconsistent with ab initio models. Unfortunately, within the community, there has long been a separation between those working on high-fidelity and low-fidelity models. In this context, multi-fidelity methods promise to become a game changer. They are based on the idea to employ a hierarchy of self-consistent models at different fidelity levels in a synergistic way to combine maximum accuracy with minimum effort. Initial applications in fusion plasma physics suggest that enormous speedups can be achieved via the use of multi-fidelity methods with respect to conventional approaches for a wide range of applications. In the present workshop, we will aim at exploring their full potential and addressing some of the challenges in state-of-the-art approaches, e.g., non-deterministic models, moving beyond moments, efficiently estimating correlations, etc., in the context of plasma physics applications.
Emily Belli
(General Atomics)
Hans-Joachim Bungartz
(Technical University Munich (TUM))
Frank Jenko
(Max Planck Institute for Plasma Physics and TU Munich)
Elizabeth Qian
(Georgia Institute of Technology)
Copyright. All Rights Reserved.